Hi David,
Thanks for your comments.
On Thu, 23 Oct 2008 09:53:00 -0700, David Brownell wrote:
> On Thursday 23 October 2008, Manuel Lauss wrote:
> > /*
> > + * LM70:
> > * The "raw" temperature read into rxbuf[] is a 16-bit signed 2's
> > * complement value. Only the MSB 11 bits (1 sign + 10 temperature
> > * bits) are meaningful; the LSB 5 bits are to be discarded.
> > @@ -80,8 +82,23 @@ static ssize_t lm70_sense_temp(struct device *dev,
> > * by 0.25. Also multiply by 1000 to represent in millidegrees
> > * Celsius.
> > * So it's equivalent to multiplying by 0.25 * 1000 = 250.
> > + *
> > + * TMP121:
> > + * 13 bits of 2's complement data, discard LSB 3 bits. Chip
> > + * transmits high byte first. Resolution 0.0625 degrees celsius.
>
> For SPI, everything is MSB first ... else insist (spi->mode & SPI_LSB_FIRST)
> is true. Please strike the comment implying it might not be MSB first.
Except that the LM70 transmits the LSB first, so it seems valuable to
underline that the TMP121 behaves differently.
> > */
> > - val = ((int)raw/32) * 250;
> > + switch (p_lm70->chip) {
> > + case LM70_CHIP_LM70:
> > + raw = (rxbuf[1] << 8) + rxbuf[0];
> > + val = ((int)raw/32) * 250;
> > + break;
> > +
> > + case LM70_CHIP_TMP121:
> > + raw = (rxbuf[0] << 8) + rxbuf[1];
> > + val = (raw / 8) * 625 / 10;
> > + break;
> > + }
> > + dev_dbg(dev, "raw=0x%x\n", raw);
> > status = sprintf(buf, "%d\n", val); /* millidegrees Celsius */
> > out:
> > mutex_unlock(&p_lm70->lock);
--
Jean Delvare
